Dec 1, 2017
Lopata Hall, Room 101
"Feasibility and Experience of FitBit -based Monitoring of Heart Failure Patients"
Adviser: Chenyang Lu
Hospital readmission rate is high for heart failure patients. Early detection of deterioration will help doctors prevent readmissions, thus reduce health care cost and provide patients with just-in-time intervention. Wearable device s (e.g., wrist bands and smart watches) provide a convenient technology for continuous outpatient monitoring. This talk presents our experience of monitoring heart failure patients using FitBit ChargeHR wristbands and preliminary results of predicting hospital readmissions. We first describe our data collection system and the clinical study involving 25 heart failure patients recently discharged from the Barnes-Jewish Hospital. We then analyze the feasibility of continuously monitoring outpatients using wristbands by analyzing the yied and latency of the health data collected through the Fitbit ChargeHR bands. Finally, we present preliminary results on using machine learning to predict readmissions based on the Fitbit data. For deterioration prediction, we compare commonly used machine learning models via leave-one-out (LOO) cross validation and 5-fold cross validation, and find logistic regression is the best model with LOO accuracy of 0.92 and 5-fold accuracy of 0.9667. We show that our wristband-based monitoring approach significantly outperforms tradition approach es , such as LACE index, in terms of predicting deterioration, which includes readmission and mortality.
"Multi-Mode Virtualization for Soft Real-Time Systems"
Adviser: Chenyang Lu
Real-time virtualization is an emerging technology for embedded systems integration and latency-sensitive cloud applications. Earlier real-time virtualization platforms require offline configuration of the scheduling parameters of virtual machines (VMs) based on their worst-case workloads, but this static approach results in pessimistic resource allocation when the workloads in the VMs change dynamically. Here, we present Multi-Mode-Xen (M2-Xen), a real-time virtualization platform for dynamic real-time systems where VMs can operate in modes with different CPU resource requirements at run-time. M2-Xen has three salient capabilities: (1) dynamic allocation of CPU resources among VMs in response to their mode changes, (2) overload avoidance at both the VM and host levels during mode transitions, and (3) fast mode transitions among different modes. M2-Xen has been implemented within Xen 4.8 using the real-time deferrable server (RTDS) scheduler. Experimental results show that M2-Xen maintains real-time performance in different modes, avoids overload during mode changes, and performs fast mode transition.